Add computation of saveLayer information to RecordingBench

Add computation of saveLayer information to RecordingBench

In (Add flag to beginRecording to request saveLayer information - https://codereview.chromium.org/721883002/) I claimed the extra recording cost would be negligible. This CL attempts to put some numbers behind that.

Committed: https://skia.googlesource.com/skia/+/80376531950b54c77d8084e6d81cd83c1592d92c

[robertphillips, 2014-11-18 19:09:45]

robertphillips@google.com changed reviewers:
+ mtklein@google.com

[mtklein, 2014-11-18 19:21:20]

LGTM

Was curious and tested this locally.  Very little perf impact on recording. 
CollectLayers was 8th hottest on the profile, lower than PathCounter, but higher
than building the BBH.

Next step is to just turn it on?

[robertphillips, 2014-11-18 19:44:12]

The CQ bit was checked by robertphillips@google.com

[commit-bot: I haz the power, 2014-11-18 19:44:38]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/741523002/1

[commit-bot: I haz the power, 2014-11-18 19:53:07]

Committed patchset #1 (id:1) as
https://skia.googlesource.com/skia/+/80376531950b54c77d8084e6d81cd83c1592d92c

[robertphillips, 2014-11-19 14:33:46]

It looks like this only reliably moved the needle for the desk_carsvg.skp case.
This skp has the most layers (179) so that isn't all that surprising.

[mtklein, 2014-11-19 14:46:05]

On 2014/11/19 14:33:46, robertphillips wrote:
> It looks like this only reliably moved the needle for the desk_carsvg.skp
case.
> This skp has the most layers (179) so that isn't all that surprising.

+1.  I'm not even sure I see a change to desk_carsvg